让AI Agent在执行中自我进化，边做边学技能。
FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills
- 执行中将成功流程编译为可复用技能，存入持久化库。
- 在ALFWorld上达85.6%准确率，比最强基线高26.4点。
- 无需训练，适配多种模型，适合构建自进化AI系统。
大型语言模型代理可通过推理时构建工作流来适应复杂任务，但单次执行后的流程通常被丢弃。现有技能库虽可复用，但多为离线构建，无法从代理自身工作流中生长。我们提出FlowEvo，一种无需训练的框架，使工作流与技能在推理时协同进化。FlowEvo将成功工作流编译为可调用技能，存入持久化知识库，并通过直接执行或作为上下文构造新工作流。它还追踪每个技能的下游效用，抑制导致负迁移的技能。使用共享GPT-4o-mini主干，在ALFWorld、HumanEval、MBPP、GSM8K和MATH-500的标准全集上，性能优于8个基线。在ALFWorld上达到85.6%准确率，较最强基线高出26.4个百分点，同时仅消耗约三分之一的令牌。在10种基础模型（7B至671B参数）上，50组模型-数据集组合中，有49组优于ExpeL。代码已开源：https://github.com/DEFENSE-SEU/FlowEvo。
原文摘要 · Abstract (English)
Large language model agents can adapt to complex tasks by constructing workflows at inference time, but procedures discovered in one episode are usually discarded after execution. Existing skill libraries provide reusable executable routines, but are typically assembled offline and do not grow from the agent's own workflows. We introduce FlowEvo, a training-free framework in which workflows and skills co-evolve at inference time. FlowEvo compiles successful workflows into callable skills, stores them in a persistent bank, and uses retrieved skills either through direct execution or as context for constructing new workflows. It also tracks each skill's downstream utility and suppresses skills that cause negative transfer. Using a shared GPT-4o-mini backbone, FlowEvo achieves the highest accuracy among 8 baselines on the full standard splits of ALFWorld, HumanEval, MBPP, GSM8K, and MATH-500. On ALFWorld, it reaches 85.6%, 26.4 points above the strongest baseline, while using roughly one third as many tokens. Across 10 base models spanning 7B to 671B parameters, FlowEvo outperforms ExpeL in 49 of 50 model-dataset comparisons. Code is available at https://github.com/DEFENSE-SEU/FlowEvo.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。